Add an anti-hallucination node in agentic workflow

[RamiAwar]

Privileged issue

• I'm @RamiAwar or he asked me directly to create an issue here.

Issue Content

Sometimes, LLM hallucinates data in the response, even when data security is enabled.

This is problematic because:

1. Confuses users and makes them think security is compromised when it's not

[shyamdurai]

Hey! Did you solve this at all?

Here is what we built - It will be open sourced early Jan 2025:

https://provably.ai/blog/introducing-proving-a-technique-to-rapidly-verify-and-trust-ai-answers

You are welcome to try it - I d love your feedback. Just fill out the form or mail me on shyam at provably dot ai.